Casting Metal Parts

October 20, 2017

Casting Metal Parts

Warning: Metal casting is dangerous. Molten metal and hot objects can cause burns. Additionally, many of the molding materials are toxic. Please seek the guidance of a trained professional before attempting anything listed here. All of our metal castings were performed under the supervision of a professional metal fabricator.

I spent the weekend having the most fun I've had in quite some time - casting metal parts from 3D printed objects. I'll share my experiences here along with what I learned.

Special thanks to Dana (Instagram: @phoenix55x) for the use of his hot-rod shop and helping us out!

In total we made five attempts at casting, two of which were successful:

Attempt #1: Sand Casting

Our ultimate goal was to make an investment casting of this brake lever that we printed with moldlay filament:

However, since investment casting destroys the part, we decided to first try a sand casting as an experiment. In a sand casting, the part is reusable since it is removed from the mold before the metal pour.

We used Green Sand for the mold material. Green sand is a reusable mixture of sand, clay (bentonite), and water. Special thanks to Queen City Clay (instagram: @queencityclay) for their advice and help in selecting the materials. We didn't use the plaster in this step - that's used later for investment casting.

We used respirators at all times when handling the materials since there were warnings on the packaging about toxicity. For the Green Sand, we mixed about 50 lbs of sand to 6 lbs of bentonite, while spraying a fine mist of water on with a spray bottle. The exact proportion is found by trial and error. A proper mixture will allow you to create a "sausage" in your hand by squeezing the material tightly. The sausage should hold together on it's own and break apart along clean lines with some resistance.

We used a two-part mold for our sand casting. We quickly discovered that our part wasn't quite suitable for this type of casting. Due to the complex geometry of this part (holes, cutouts and lack of draft) we had to tape over the holes and cutouts to make the part removable.

We placed the mold frame on a flat surface and then elevated the printed part on blocks of wood so that the centerline of the part was in the same plane as the dividing line of the two mold halves. Next we began to add the sand, sifting the first layer to ensure that it was free of chunks.


We then filled the mold frame with sand, tamping the sand down tightly with a large block of wood. After filling with sand, the part stuck up half-way out of the bottom mold. Next, we flipped the (now filled) bottom frame over, placed the second frame on top,  and sprinkled baby powder on the bottom half to prevent the two halves from sticking together when we would separate them later. Then we filled the second frame with sand as well:

After we filled the mold frame with tightly packed sand, we *carefully* separated the mold halves and removed the part. The molds didn't separate cleanly and some of the sand broke off in the process - likely due to the rough surface of the tape we used to seal off the holes in the part.

Time for the pour! We fired up our $70 furnace, which consisted of a 5 gallon bucket filled with fireproof concrete and powered by a propane tank.

The results were mediocre yet encouraging for our first attempt. The handle came out nicely but the rest of the part was mangled. Molten metal had somehow managed to escape from between the two mold halves.

From this we learned that it's important to incorporate draft (sloped sides) into the printed part to allow the part to be removed cleanly from the sand without damaging the fragile mold. We also learned that two-piece molds are a little trickier that we originally expected due to the potential for molten metal to seep out into the middle of the two halves.

Attempt #2 - Investment Casting

Next we tried to investment cast the part. Investment casting involves creating a plaster mold around the part, melting the part of the mold, and then pouring molten metal into the space left behind. The plaster mold is then broken away, revealing the part - or so we thought...

Our investment mixture was a 2:1 ratio of fine sand and plaster, with water added to until we got a yogurt-like texture. We placed the part in the bottom of a pitcher and poured in the investment.

After it dried (about 30 mins) we flipped the part over and carved a small hole.

Our next step was to melt out the moldlay filament. We placed the mold upside-down on our furnace and heated it for 2 hours, but only about 3 inches of the part melted out.

Perhaps we could have achieved better melt-out inside of an oven, but we didn't have the time, so we decided to move on to more sand-castings.

Attempt #3 - Multiple Sand Castings

We learned from our previous attempts that to increase our chances of getting good castings, we should: 1) use sand casting with single-piece molds, and, 2) add draft to all sides of our parts. Based on these findings we created three new parts, designed for a straight-pull from a single mold, with at least 5 degrees of draft on all surfaces. We printed these parts from PLA, instead of moldlay, because we don't need to melt them out of the mold.

Next, we created a single-piece mold containing all three printed parts from Green Sand. The left part separated cleanly from the mold, leaving all features intact - the MakerGear logo on this part had a generous 10 degrees of draft. The middle part damaged the mold when we pulled it out - a lot of sand stuck to the printed part near the makergear logo, likely because this feature had only 2 degrees of draft. The right part also damaged the mold slightly during removal. While this part had 10 degrees of draft on the 'MAKERGEAR' lettering, the detail was much finer, which is likely why some of the sand pulled out with the printed part.

Now for the fun part - pouring the molten metal!

Our pour was a little messy and required quite a lot of post-processing with a band saw and bench sander (an 80 grit belt worked best), but the results were pretty good!

All features on the three-pin link came out pretty well. The MakerGear logo is a little shallow because the metal didn't quite flow into the grooves.

The wrench casting turned out fairly well also. The MakerGear logo is somewhat visible even though the most of the sand was removed from this feature when the printed part was taken out of the mold. Next time we'll make the letters thicker and with even more draft (currently 10 degrees). We still have some post-processing to do on this part.

We decided to pour the medallion even though the mold was largely damaged during the extraction of the printed part. Even so, we were surprised that this casting turned out as well as it did. Next time we'll add more draft to the logo to make the printed part easier to remove from the mold.

Summary - Lessons Learned

The casting process was trickier than we expected, requiring three attempts to obtain decent parts. We obtained our best results using:

  • Sand casting with Green Sand
  • A one-piece mold
  • Printed parts designed with at least 5 degrees of draft on large features and 10 degrees of draft on small features
  • Printed parts with large features to allow the metal to flow freely

For the Future...

Next, we plan to place the investment into an oven for a few hours to try to fully melt the moldlay filament out of the mold. We also want to try to make our letters thicker and add even more draft - perhaps 20 degrees. Given the success we had with sand casting using one-piece molds, we'd also like to experiment more with two-piece molds.